In ddsjoberg/gtsummary: Presentation-Ready Data Summary and Analytic Result Tables

knitr::opts_chunk$set(echo = TRUE)

library(gtsummary)
library(survival)
library(lme4)

options(gtsummary.as_gt.addl_cmds = "gt::tab_options(table.font.size = 'small', data_row.padding = gt::px(1))")

Build models, then do tbl_regression and examine results. Do quick visual check against broom::tidy().

Linear Model

# 
mod_lm <- lm(hp ~ am, data = mtcars)
tbl_regression(mod_lm)

broom::tidy(mod_lm, conf.int = TRUE)

Survival Model

mod_survreg <- survreg(Surv(time, status) ~ age + ph.ecog, data = lung)
tbl_regression(mod_survreg)
broom::tidy(mod_survreg, conf.int = TRUE)

Logistic Model

mod_logistic <- glm(response ~ age + stage, trial, family = binomial)
tbl_regression(mod_logistic, exponentiate = TRUE)
broom::tidy(mod_logistic, exponentiate = TRUE, conf.int = TRUE)

Poisson Model

mod_poisson <- glm(count ~ age + trt,
  trial %>% dplyr::mutate(count = sample.int(20, size = nrow(trial), replace = TRUE)),
  family = poisson
)
tbl_regression(mod_poisson, exponentiate = TRUE)
broom::tidy(mod_poisson, exponentiate = TRUE, conf.int = TRUE)

Linear Mixed Model

mod_lmer <- lmer(Reaction ~ Days + (Days | Subject), sleepstudy)
tbl_regression(mod_lmer)
broom.mixed::tidy(mod_lmer)

Generalized Linear Mixed Model

mod_glmer <- glmer(am ~ hp + factor(cyl) + (1 | gear), mtcars, family = binomial)

tbl_regression(mod_glmer, exponentiate = TRUE)
broom.mixed::tidy(mod_glmer, exponentiate = TRUE, conf.int = TRUE)

tbl_lme4 <- tbl_regression(mod_glmer, exponentiate = TRUE,
                               conf.level = 0.90)

a <- coef(mod_glmer)[[1]] %>% {.[1, 2:ncol(.)]} %>% purrr::map_dbl(exp) 

b <- tbl_lme4$table_body %>% 
  dplyr::pull(estimate) %>% 
  na.omit()

all.equal(unname(a), as.vector(b))

Linear Model with Interaction

mod_lm_interaction <- lm(age ~ trt * grade * response, data = trial)
tbl_regression(mod_lm_interaction)
broom::tidy(mod_lm_interaction, conf.int = TRUE)

Cox PH with Hmisc Labels

lung2 <- lung
Hmisc::label(lung2$sex) <- "Gender"
Hmisc::label(lung2$age) <- "AGE"
cox_hmisclbl <- coxph(Surv(time, status) ~ age + sex, data = lung2)
tbl_regression(cox_hmisclbl, exponentiate = TRUE)
broom::tidy(cox_hmisclbl, exponentiate = TRUE, conf.int = TRUE)